Objective To investigate the optimization of neonatal brain MR imaging with knee phased-array coil. Methods Forty-five newborns participated in the study and were scanned with a knee phased-array coil. Imaging techniques, including T1- weighted imaging, T2-weighted imaging, T2 FLAIR imaging, and diffusion weighted imaging at the axial orientation, and T1-weighted imaging at the sagittal orientation, were respectively conducted. Twenty newborns were scanned with the conventional imaging parameters, and the other twenty-three were scanned with the optimized imaging parameters. Signal-to noise ratio (SNR) and image quality for clinical diagnosis were compared between paired scans. Results All data was successfully acquired. No statistically significant difference of SNR was observed between the conventional and the optimized imaging parameters (P-value > 0.05). In addition, a Chi-square test suggested no significant difference of image quality for clinical diagnosis between the two sets of imaging parameters (P-value > 0.05). The time scanned with the optimized imaging parameters was less than that scanned with the conventional imaging parameters. Conclusion With the advantages of halving scan time and specific absorption rate, the optimized neonatal brain MR imaging parameters were superior to the conventional ones, and were more suitable for neonatal brain MR examination with knee phased-array coil.